Machine Learning Enabled Visual Servoing with Dedicated Hardware Acceleration

2. The system of claim 1, wherein a target destination for the workpiece object is a receiving device comprising an aperture into which the workpiece object is to be inserted, wherein the desired pose configuration is retrieved from the receiving device.

3. The system of claim 2, wherein the machine learning model is configured to extract features of the aperture from the visual feedback information.

5. The system of claim 1, wherein the workpiece object is unknown to the system controller and the machine learning model determines pose configuration error based on training of known objects.

6. The system of claim 1, wherein the hardware accelerator is configured with an architecture optimized for high bandwidth but low power operation.

7. The system of claim 1, wherein the hardware accelerator is configured with an architecture that accelerates by minimizing data transfer within built-in memory or by accelerating matrix multiplication.

8. The system of claim 1, wherein the machine learning model is configured as a deep neural network, a support vector machine, a random forest, or a hidden Markov model.

9. The system of claim 1, wherein the system controller comprises a plurality of technology modules, with each technology module dedicated to a different task for the controlling the robotic device.

10. The system of claim 1, wherein visual sensor inputs to each technology module are of a different type or modality of high speed inputs.

11. The system of claim 1, wherein the technology module receives visual feedback information from a plurality of visual sensors to provide depth perception.

12. The system of claim 1, wherein the one or more sensors include at least one 3D point cloud camera.

14. The system of claim 1, wherein the servo control module is programmed to set limits to values of the servo control signal to avoid a discovered obstacle condition.